Benedita Malheiro
[email protected]
Jeremy Foss
[email protected]
Juan Carlos Burguillo
[email protected]
Ana Peleteiro
[email protected]
Fernando Mikic
[email protected]
Introduction
Framework for the dynamic personalisation of media content.
Players: Media content producers, media service distributors, and viewers.
Two sorts of video objects: User-selected stream and external streams.
Profiling of the viewers is continuously performed by the system.
The identification of the potential candidate objects is performed by an
agent-based brokerage platform.
Introduction
Background
COAST: Content-centric network overlay architecture linking content
sources to content consumers, including content-aware retrieval, delivery
and streaming.
CAM4HOME: Metadata enabled content delivery framework to support
end users and commercial content providers to create and deliver rich
multimedia experiences.
ACDC: User-aware content-delivery architecture for content navigation
and personalisation based on semantics and clouding.
OMWeb: Personalised media objects constructed on the fly from
distributed components and according to user preferences.
MPEG-7: Standard developed by the Moving Picture Experts Group
(MPEG) for the structural and semantic description of multimedia content.
MPEG-7 based ontologies: Harmony, SmartWeb, Boemie, DS-MIRF,
Rhizomik, aceMedia Project RDFS and Core Ontology for Multimedia.
To adopt an MPEG-7 based ontology for the representation of all media
content.
Content-based Filtering (CBF): It uses the description of the resource and
the user’s interests to provide recommendations.
Collaborative Filtering (CF): It usually considers the comparison of ratings,
provided by the users to the resources, with other similar users
(concerning their profile).
Collaborative Tagging (CT): It allows users to describe contents by means
of tags and to share such description.
Case-based Reasoning (CBR): It focuses on inexact reasoning by a
similarity measurement among cases.
E-brokers: Electronic intermediaries that facilitate exchanges between
buyers (consumers) and sellers (producers) by meeting the needs of both
parties.
E-brokerages are still a relative new and poorly understood type of
business.
The existing so-called media e-brokerage systems are mainly for resource
allocation and not for media content delivery.
Frame-based overlays (objects are overlays images) and Object-based
video (supported by MPEG-4 extensions).
Digital Media Broadcasting (DMB): It usually considers the comparison of
ratings, provided by the users to the resources, with other similar users
(concerning their profile).
Transmitted video objects are multiplexed via the Delivery Multimedia Integration
Framework (DMIF).
Objects are related to each other in their video scene and in their temporal relationships
by the Binary Interchange Format (BIFS) or through the Lightweight Application Scene
Representation (LASeR).
The user-end equipment will reassemble the objects into the desired scene for playback
to the user interface.
Introduction
Background
Network Architecture
Potential network implementation 
Integration server:
It receives the source video file from the
head-end.
It will request and retrieve the externally
acquired video objects from third party
video object servers.
Selecting video:
The source metadata requirements.
The external video object metadata description.
The profile of the viewer.
Re-multiplexing and streaming.
Introduction
Background
Network Architecture
System Overview
Generic architecture 
Video objects are MPEG-4 instances
annotated in an MPEG-7 based OWL
ontology.
Enterprise agents: Producers and distributors of media content
modelled by autonomous intelligent agents.
They can publish, update and remove their service descriptions – metadata
descriptions of the objects they hold or seek to insert in the viewer stream.
Any entity can discover, download and interact with any service (agent) automatically.
Viewer profiling 
To find the best possible objects to
insert within the mainstream:
Users’ profiles as CBR cases containing information (personal and semantic data, and
tag cloud).
Filtering techniques enhaced with collaborative tagging to search for recommended
objects.
P2P scheme to perform distributed search.
Brokerage platform (2 layers) 
Top layer  Agents representing the
producers and distributors (enterprise
agents and market profiler agent).
Bottom layer  Agents constituting the marketplace, which is
governed by the market agent and populated by delegate agents.
Introduction
Background
Network Architecture
System Overview
Current Status
Framework is under development.
Brokerage platform  JADE.
Ontology  OWL.
Media components  MPEG-7 based OWL ontology.
Currently types of markets  Iterated Contract Net negotiation protocol.
Introduction
Background
Network Architecture
System Overview
Current Status
Conclusion
Ambitious solution to the dynamic media content personalisation
challenge.
Scalable market model to cover the global media industry.
Based on existing standards.
Design concept  Open, modular, and open source and Java based.
Semantic Web (ontology-based) approach for the knowledge
representation intends to contribute to promote the interoperability
with other systems and to allow future expansions.
Benedita Malheiro
[email protected]
Jeremy Foss
[email protected]
Juan Carlos Burguillo
[email protected]
Ana Peleteiro
[email protected]
Fernando Mikic
[email protected]